In this paper we assess the performance of farmers adopting an improved sorghum technology package in the Maradi region of Niger. A propensity score matching method is used to address self-selection bias into the program. First, we estimate a propensity model to participate in the extension program, examine factors affecting the participation, and assess the average adoption effect on participants by testing for productivity difference between adopters and non-adopters. Secondly, we estimate a stochastic production technology frontier to compare technical efficiency of farmers in the extension program. We test for returns to scale, examine factors affecting technical efficiency of participants, and compare technical efficiency scores of participants based on their compliance to program recommendations. Participants in the extension program are older, have less farming experience, and operate on larger farm sizes. After controlling for bias, there is very little change in the yield differences, which in both cases are substantial. There is some evidence for greater productivity of the smaller size producers.Technical efficiency of participants is increasing overtime, younger participants are technically more efficient and farming experience increases significantly participants’ technical efficiency. Good followers of the fertilization technique recommendation are much more technically efficient.